Timestamps are in GMT/BST.
[0:00] * Arfrever (~Arfrever@apache/committer/Arfrever) Quit (Quit: Ex??re)
[0:20] * stakkars (~tismer@p5DDB5E98.dip.t-dialin.net) Quit (Quit: stakkars)
[0:26] * stakkars (~tismer@p5DDB5E98.dip.t-dialin.net) has joined #jython
[1:09] * powerje_ is now known as powerje
[5:54] * agronholm (~demigod@nblzone-211-37.nblnetworks.fi) Quit (Remote host closed the connection)
[5:56] * agronholm (~demigod@nblzone-211-37.nblnetworks.fi) has joined #jython
[13:22] * agronholm (~demigod@nblzone-211-37.nblnetworks.fi) Quit (Remote host closed the connection)
[13:22] * agronholm (~demigod@nblzone-211-37.nblnetworks.fi) has joined #jython
[15:41] * shashank (~shashank@65-101-219-224.hlrn.qwest.net) has joined #jython
[15:49] * shashank (~shashank@65-101-219-224.hlrn.qwest.net) Quit (Quit: Leaving.)
[15:51] * shashank (~shashank@65-101-219-224.hlrn.qwest.net) has joined #jython
[16:01] * Trundle (~andy@python/site-packages/trundle) has joined #jython
[17:01] * Arfrever (~Arfrever@apache/committer/Arfrever) has joined #jython
[17:24] * irmen (~irmen@razorvine.xs4all.nl) Quit (Remote host closed the connection)
[18:12] * jabley (u2487@gateway/web/irccloud.com/x-tmsbhbubnirdtjen) Quit (Quit: Connection closed for inactivity)
[19:22] * stakkars_ (~tismer@p5DCE2768.dip.t-dialin.net) has joined #jython
[19:38] * stakkars_ (~tismer@p5DCE2768.dip.t-dialin.net) Quit (Quit: schnarch)
[19:51] <agronholm> py4j is awesome...I can use JasperReports from my CPython app without writing a single line of Java code (or using a heavyweight solution like jython)
[20:30] <pjenvey> agronholm - no idea on that pdb problem, never heard of something like that
[20:30] <agronholm> curious thing
[20:30] <agronholm> from sqlalchemy.orm.util import class_mapper
[20:30] <agronholm> takes 3.24 seconds on jython on my laptop
[20:31] <agronholm> even when all the modules are compiled
[20:31] <agronholm> over 5 seconds otherwise
[20:31] <agronholm> that's just...awful
[20:31] <agronholm> 0.08s on CPython
[20:33] <agronholm> pjenvey: I tried with the latest dev version too
[20:33] <agronholm> had hoped it'd magically improve things
[20:38] <pjenvey> agronholm - is it that bad after the disk is cached
[20:38] <agronholm> yeah
[20:38] <agronholm> I'm running on an SSD anyway
[20:39] <pjenvey> bytecode verification is unfortunately pretty heavy
[20:39] <pjenvey> i tried profiling it once, that was most of the time
[20:39] <pjenvey> it almost seems worse than typical .class files
[20:39] <agronholm> why don't native java programs suffer from the same problem then
[20:39] <pjenvey> i wonder if our generated bytecode just takes longer to verify than stuff output by javac or somemthing
[20:39] <pjenvey> i don't know, it would take ssome investigation
[20:40] <agronholm> maybe worth looking into
[20:42] <agronholm> also, I understand jruby does not have this problem...or?
[20:43] <agronholm> I heard it's faster than the mainline ruby
[20:49] <pjenvey> that's partly because jruby (last time I checked anyway) doesn't byte compile to disk
[20:49] <pjenvey> it parses .rb code everytime
[20:50] <pjenvey> and then has an AST walking interpreter which they've done a fair amount of work on to make fast
[20:50] <agronholm> how can that be faster
[20:50] <pjenvey> yea it sounds crazy that it would be
[20:50] <pjenvey> but it is
[20:51] <agronholm> was jimbaker working on something similar for jython?
[20:51] <pjenvey> so we have an python bytecode interpreter that isn't enabled by default
[20:51] <pjenvey> jim wrote it
[20:51] <agronholm> yeah that
[20:51] <agronholm> does it work?
[20:51] <pjenvey> so it's pretty much a big switch() interpreter
[20:51] <pjenvey> which the JVM is not very good at optimizing unfortunately
[20:52] <pjenvey> so our interpreter hasn't been optimized further to counter that, it's slow
[20:52] <pjenvey> it's kind of difficult making a faster byte code interpreter on the JVM, our recommendations from hotspot engineers is to actually use nested if else for common bytecodes
[20:53] <pjenvey> it's a lot of work to figure out the right formula there
[20:53] <agronholm> I bet
[20:53] <pjenvey> however in Java 7 potentially we can build an interpreter with MethodHandles
[20:53] <pjenvey> and that may not be fast either on Java 7 u0/1/2, but hotspot guys could probably take it and make it faster for us
[20:54] <pjenvey> kind of sucks we have a lot of work here but nobody has had the time
[20:55] <pjenvey> Oracle has funded a new impl of JAvascript on the JVM, called 'naghorn' IIRC
[20:56] <pjenvey> maybe they are investigating interpreter perf, i don't know. it's not even open source yet
[20:56] <pjenvey> i think remi forax' phpreboot has a fairly fast interpreter
[21:59] * shashank (~shashank@65-101-219-224.hlrn.qwest.net) Quit (Ping timeout: 258 seconds)
[22:14] * stakkars_ (~tismer@i59F6D818.versanet.de) has joined #jython
[22:16] * stakkars_ (~tismer@i59F6D818.versanet.de) Quit (Client Quit)
[22:21] * Trundle (~andy@python/site-packages/trundle) Quit (Remote host closed the connection)
[22:24] * stakkars_ (~tismer@i59F6D818.versanet.de) has joined #jython
[22:50] <pjenvey> oh crap i almost forgot
[22:50] <pjenvey> im pretty sure Alan has screwed up the 2.5 branch
[22:58] <agronholm> ?
[22:58] <agronholm> ah yeah the 2.5 branch thinks it's 2.6
[22:59] <agronholm> I had some difficulties with it
[23:19] <pjenvey> yea
[23:19] <pjenvey> fixing it is somewaht easy
[23:19] * pjenvey thinking the best way to deal with the crappy 2.5 branch alan created
[23:20] <pjenvey> close it might be odd
[23:20] <pjenvey> closing it
[23:21] <pjenvey> think i have to close it..
[23:22] <pjenvey> merging it with the correct 2.5... might be even wackyer
[23:25] * stakkars_ (~tismer@i59F6D818.versanet.de) Quit (Ping timeout: 260 seconds)
[23:32] * shashank (~shashank@65-101-219-224.hlrn.qwest.net) has joined #jython
[23:35] * stakkars_ (~tismer@p5DC4677E.dip0.t-ipconnect.de) has joined #jython
These logs were automatically created by JythonLogBot_ on irc.freenode.net using a slightly modified version of the Java IRC LogBot (github).